arXiv:1603.

08948v1 [math-ph] 29 Mar 2016

The stationary measure for diagonal quantum walk with one defect

Takako Endoa
Department of Physics, Ochanomizu University
2-1-1 Ohtsuka, Bunkyo, Tokyo, 112-0012, Japan
LERMA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universit
es, UPMC Univ. Paris 6, UMR 8112, F-75014
4 place Jussieu 75005 Paris, France
Hikari Kawaib
Department of Appllied Mathematics, Faculty of Engineering, Yokohama National University
Norio Konnoc
Department of Appllied Mathematics, Faculty of Engineering, Yokohama National University
79-5 Tokiwadai, Hodogaya, Yokohama, 240-8501, Japan

Abstract
This study is motivated by the previous work [14]. We treat 3 types of the one-dimensional
quantum walks (QWs), whose time evolutions are described by diagonal unitary matrix, and diagonal
unitary matrices with one defect. In this paper, we call the QW defined by diagonal unitary matrices,
the diagonal QW, and we consider the stationary distributions of generally 2-state diagonal QW
with one defect, 3-state space-homogeneous diagonal QW, and 3-state diagonal QW with one defect.
One of the purposes of our study is to characterize the QWs by the stationary measure, which may
lead to answer the basic and natural question, What the stationary measure is for one-dimensional
QWs ? . In order to analyze the stationary distribution, we focus on the corresponding eigenvalue
problems and the definition of the stationary measure.

1 Introduction
Quantum walk (QW) is a quantum mechanical version of classical random walk, whose time evolution
of QWs are defined by unitary evolutions of probability amplitudes. For its characteristic properties,
QW has attracted much interest of various fields in the quantum scale, such as quantum algorithms
[1, 18] and quantum physics [8, 10, 15, 22]. Many reviews and books on QWs have been published so far
[2, 9, 11, 16, 19, 20]. Owing to the wealth applications, it is beneficial to study QWs both theoretically
and experimentally. In recent years, QWs have also been implemented experimentally by several kinds
of materials, such as trapped ions [21] and photons [17]. However, because of its difficulties for QWs
to implement the state after many steps, we have not succeeded to see experimentally the behavior in
aendo.takako@ocha.ac.jp

(e-mail of the corresponding author)

b kawai-hikari-dy@ynu.jp
c konno@ynu.ac.jp

the long time-limit. In addition, for the quantum nature, it is hard for us to intuitively assimilate the
properties.
Recently, to derive measures from QWs has become one of the hottest topics in the theoretical study
of QWs [3, 5, 6, 7, 12, 14]. Especially, the properties of the stationary measures of the two-state QWs in
one dimension have been gradually studied in the immediate past [3, 5, 7, 12, 14]. As is well known, the
stationary measure of the Markov chain has been deeply studied, however, the fruits of study for QW has
not yet been given. This is one of the motivations of our study. In this paper, we focus on the 2-state
and 3-state QWs in one dimension whose time evolutions are generally expressed by a diagonal unitary
matrix or diagonal unitary matrices with one defect.
Now we explain concisely the previous studies of stationary measures. In 2013, Konno et al. [13] gave
a stationary measure of the QW with one defect whose quantum coin has a phase at x = 0, and Konno [12]
gave the uniform measure as a stationary measure of the one-dimensional QWs in one dimension. Then
Endo et al. [5] got a stationary measure of the QW with one defect whose quantum coins are defined by
the Hadamard matrix at x 6= 0 and the rotation matrix at x = 0. Endo and Konno [3] derived a stationary
measure of the Wojcik model in 2014, and then Endo et al. [7] obtained a stationary measure of the
two-phase QW. Konno and Takei [14] showed that the set of the stationary measures of the QW except
U is diagonal, contains non-uniform stationary measure and they showed that any stationary measure is
uniform for diagonal matrices. Moreover, they proved that the set of the stationary measures contains
uniform measure for the QW in general.
The organization of this paper is as follows. In the next section, we give a description of n-state discretetime QW (DTQW) in one dimension. We consider the stationary measures of the 2-state diagonal QW
with one defect in Section 3, and then, we study 3-state case in Section 4.
2 Description of n-state DTQW in one dimension
As a quantum version of classical random walk with discrete-time, discrete-time QW (DTQW) has been
intensively studied since the beginning of 2000. In this section, we explain briefly how the n-state DTQW
is generally described, where n N. The total space is defined by a Hilbert space H:
H = HP HC ,
where HP is spanned by {|xi; x Z}, called a position Hilbert space, and HC is a coin Hilbert space,
spanned by {|Ji; J {L, R}}, where Z is the set of integers. The time evolution of the walk is described
by a set of n n unitary matrices {Ux }xZ on HC , where

x11 x12 x1n

..
x21 . . .
.

Ux =
.
.. (xij C, i, j N),
..
..
.
.
xn1 xnn
which is called the quantum coin. Note that the subscript x Z represents the position of the walker. To
define the time evolution, let us divide the matrix Ux into n parts as follows:
Ux = V1 + V2 + + Vn ,
where

x11

0
V1 =
.
..
0

x12
..
.

x1n

..

0
..
.

0
x21

V2 = .
..
0

x22

..
.

0
0
..

x2n
.
.. , , Vn =

.
0
0
xn1

..
.

..

xn2

As the general setting of DTQW, the inner state of the walker consists of n chiralities:
|W1 i = T[1, 0, , 0]n , |W2 i = T[0, 1, , 0]n , , |Wn i = T[0, 0, , 1]n .
At each time step, the walker steps according to the chirality as follows:

0
..
.
.

0
xnn

1. Case of |W1 i:
if n is odd, then the walker moves to the left with (n 1)/2 steps.
if n is even, then the walker moves to the left with n/2 steps.
2. Case of |W2 i:
if n is odd, then the walker moves to the left with (n 3)/2 steps.
if n is even, then the walker moves to the left with (n 2)/2 steps.
3. Case of |Wn1 i:
if n is odd, then the walker moves to the right with (n 3)/2 steps.
if n is even, then the walker moves to the right with (n 2)/2 steps.
4. Case of |Wn i:
if n is odd, then the walker moves to the right with (n 1)/2 steps.
if n is even, then the walker moves to the right with n/2 steps.
Furthermore, if n is odd and the walker has the chirality |W(n1)/2 i, then, the walker does not move. The
walker at time t and position x has a coin state expressed by the n-dimensional vector;
W2
Wn
n
1
t (x) = T[W
t (x), t (x), , t (x)] C

and the time evolution is determined by the recurrence formula:

if n is odd, we have
t (x) = V1 t1 (x+(n1)/2)+V2 t1 (x+(n3)/2)+ +Vn1 t1 (x(n3)/2)+Vn t1 (x(n1)/2).
if n is even, we have
t (x) = V1 t1 (x+ n/2)+ V2 t1 (x+ (n 2)/2)+ + Vn1 t1 (x (n 2)/2)+ Vn t1 (x n/2).
Now let

W1 W1
1
t (1)
t (0)
W
t (1)

..
..
..
n Z
t = T ,
, (C )
,
,
.
.
.

and

U (s)

n
W
t (1)

..
.
O
Q2
O
O
O
..
.

..

= SUx =

..
.
P1
O
Q1
O
O
..
.

n
W
t (0)

..
.
O
P0
O
Q0
O
..
.

..
.
O
O
P1
O
Q1
..
.

n
W
t (1)

0
O

. ..
O
.
with O = ..
P2
0

..
.

where T means the transposed operation and S is the standard shift operator:
X
S=
(|xihx + 1| |LihL| + (|xihx 1| |RihR|).
x

0
.. ,
.
0

The unitary matrix U (s) is a time evolution operator on H, that is, the state of the walker at
position x and time t can be defined by t = (U (s) )t 0 .
Now we prefer to introduce a map t : Z [0, 1]:
Wn
2
2
1
t (x) = kt (x)k2 = |W
t (x)| + + |t (x)|

f or x Z.

Assuming k0 k2 = 1, t becomes the probability measure, and we can define the random variable Xt
from t , that is, the walker can be observed at position x and time t with the probability
P (Xt = x) = t .
Here we define the stationary measure. Put R+ = [0, ), and we set a map : (Cn )Z RZ+ such that for

W1 W1
W1

(1)
(0)
(1)

..
..
..
n Z
= T ,
,
,
, (C ) ,
.
.
.
Wn (1)

Wn (0)

Wn (1)

we set the measure by

: Z R+ such that (x) = ((x)) = |W1 (x)|2 + + |Wn (x)|2 (x Z)

where j (x) (j = W1 , , Wn ) is called the stationary amplitude. Now let
s = {(0 ) RZ+ : there exists 0 such that ((U (s) )t 0 ) = (0 ) f or any t 0},

(1)

and we call the element of s , the stationary measure of the QW. Now we consider the eigenvalue problem
of the QW
U (s) () = ()

( C).

(2)

Since the unitarity of U (s) , we have || = 1, and therefore we see (() ) Ms .

3 Case 1: 2-state diagonal QW with one defect
In this section, we study the DTQW whose quantum coins are given by the set of diagonal unitary matrices
with one defect at the origin:

{Ux } =

(

ei
0

0
() ei

,
x=1,2,

a
c

b
d

x=0

(3)

where [0, 2), a, b, c, d C, and () C with | () | = 1. Note that Eq. (3) is defined by the
double sign. We assign the two different quantum coins to the positive and negative parts, respectively.
For the unitarity, we have |a|2 + |c|2 = |b|2 + |d|2 = 1, ab + cd = 0, c = b, and d = a, where C
is the determinant of U0 with | |2 = 1. Here we show the stationary measure obtained by using the
Splitted generating function method (the SGF method) developed in the previous studies [3, 7, 13]. The
detail of the derivation of Theorem 1 is given in Section 3.1.
Theorem 1 The stationary measure of the QW is written by

(1 + |b|2 )||2 + |a|2 ||2 2(ab) (x 1)

(x) =
,
||2 + ||2
(x = 0)

|a|2 ||2 + (1 + |b|2 )||2 + 2(ab) (x 1)

where = L (0), = R (0),p

and (x) is the real part of x (x C). Here the eigenvalues which give the
stationary measure are = () .

We see that the stationary measure does not depend on the two different quantum coins, that is, if
+ = , then, we get the same result. We should remark that the stationary measure does not have an
exponential decay for the position, which is in remarkable contrast to the QWs we treated in our previous
studies [3, 7]. In addition, the stationary measure is generally non-uniform for the position, however, if
b = 0, that is, the defect is defined by the diagonal unitary matrix, or if
||2 ||2 2

(ab)
= 0,
|b|2

then, the stationary measure becomes unitary measure. Here one of the interesting future problems is to
elucidate the whole picture of the set of the stationary measure.
3.1 Proof of Theorem 1
Taking advantage of the SGF method, let us solve the eigenvalue problem
U (s) = ,

(4)

where C with || = 1. Rewriting the eigenvalue problem, we have

(x) = Px+1 (x + 1) + Qx1 (x 1).
Now Eq. (5) can be expressed according to each position as follows:
1. Case of x = 2, 3, .
L (x) = ei L (x + 1),
R (x) = () ei R (x 1).
2. Case of x = 1.
L (1) = ei+ L (2),
R (1) = cL (0) + dR (0).
3. Case of x = 0.
L (0) = e+ L (1),
R (0) = () ei R (1).
4. Case of x = 1.
L (1) = aL (0) + bR (0),
R (1) = () ei R (2).

(5)

Here we introduce the generating functions of j (x) (j = L, R) as follows:

j
f+
(z) =

j (x)z x ,

j
f
(z) =

x=1

j (x)z x .

(6)

x=1

Then we obtain
Lemma 1 Put

 L

ei
(z)
0
, f (z) = f
A = z
,
R
f
(z)
0
() ei z





(a + b)z 1
a+ (z) =
, a (z) =
.
(c + d)z

Then,
A f (z) = a (z)

(7)

hold.
Noting
det A =
()

()

we chose s , l



ei
e2i
2
()
z2
,
(
+

)z
+
()
()

()
ei z

C satisfying
det A =

()

()

and |s | 1 |l

(8)

()
()
(z s() )(z l )
ei z
() ()

(9)

= (() )1 e2i .

|. Here Eqs. (8) and (9) give s l

L
R
From now on, let us derive f
(z) and f
(z) by solving Eq. (7) in Lemma 1.
L
1. Case of f+
(z). Eq. (7) gives
L
f+
(z)

(+)
= i+
det A
e



ei+
z (+) .

Putting s = ((+) )1 ei+ , we have

L
f+
(z) =

l1 z
(+) 1
) z

1 (l

(+) 1

= (l

(+) 2 2

z + (l

z + .

Hence we see
L
f+
(z) =

(+) x x

(l

x=1

z =

((+) e2i+ )x (s(+) )x z x .

x=1

(10)

L
Equation (10) and the definition of f+
(z) imply

L (x) = ((+) e2i+ )x (s(+) )x (x = 1, 2, ),

(11)

s(+) = ((+) )1 ei+ .

(12)

where

(+)

R
2. Case of f+
(z). Putting s

= 1 ei+ , we have from Eq. (7)

R
f+
(z) =

((+) )1 ei+ (c + d)z

(+)

z l

(+) 1

((+) )1 ei+ (c + d)(l

1

(+)
(l )1 z

= ((+) )1 ei+ (c + d)
= ((+) )1 ei+ (c + d)

x=1

(+) 1

((l

z)x

((+) e2i+ s(+) z)x .

(13)

x=1

R
Equation (13) and the definition of f+
(z) give

R (x) = ((+) )1 ei+ (c + d)((+) e2i+ s(+) )x (x = 1, 2, ),

(14)

s(+) = 1 ei+ .

(15)

where

() 1

L
3. Case of f
(z). Putting (l

= 1 () ei , Eq. (7) gives

()

L
f
(z) =

() 1 1

(() )1 ei (a + b)(s )1 (l
z 1

()
(s )1

Hence we have
L
f
(z) =

=
=

() 1 1

(() )1 ei (a + b)(l

()
1 s z 1
()
ei (a + b)s z 1
()
1 s z 1

X
(s() )x z x .
ei (a + b)
x=1

(16)
(17)
(18)

L
Equation (18) and the definition of f
(z) yield

L (x) = ei (a + b)(s() )x (x = 1, 2, ),
where
s() = 1 ei .

(19)

() 1

R
4. Case of f
(z). Putting (l

= ei , we have from Eq. (7)

R
f
(z) =

((s() )1 z)x .

(20)

x=1
R
Therefore, Eq. (20) and the definition of f
(z) give

R (x) = (s() )x (x = 1, 2, ),
where
s() = (() )1 ei .

(21)

As a result, we obtain
"
#
(+) 2i+ (+) x

)
(
e

(+)

((+) )1 (c + d)ei+ ((+) e2i+ s )x

 

(x) =

#
"

i () |x|

(a
+
b)e
(
)

()

(s )|x|

(x 1)

(x = 0)

(22)

(x 1)

Moreover, 4 expressions of s , that is, Eqs. (12), (15), (19), and (21) suggest
s(+) = ((+) )1 ei+ = 1 ei+
and
s() = 1 ei = (() )1 ei .
p
()
Therefore, we see = () and |s | = 1, by choosing the sign in a suitable way. Now the proof of
Theorem 1 is completed.
4 Case 2: 3-state case.
4.1 3-state diagonal QW with one defect
In this subsection, we study the DTQW whose quantum coins are given by the set of diagonal unitary
matrices with one defect at the origin:

ei
Ux = 0

0
ei
0

0
a b
0
, d e
g h
() x=1,2,

f
,

i x=0

(23)

where [0, 2), a, b, c, d, e, f, g, h, i C, and () C with | () | = 1. Here Eq. (24) is defined by

the double sign. We give the two different quantum coins in the positive and negative parts, respectively.
Here we show the stationary measure obtained by using the SGF method [3, 7, 13].

Theorem 2 The stationary measure of the QW is written by

(1 + |g|2 )||2 + |h|2 ||2 + |i|2 ||2 + 2(gh + gi + gi) (x 1)

||2 + ||2 + ||2

(x = 0)
(x) =
,

(x 1)
||2 + (1 + |c|2 )||2 + |b|2 ||2 + 2(ab + ac + bc)

where = L (0), = R (0), = L (1), = R (1), and (x) means the real
p part of x (x C).
Note that the eigenvalues which contribute to the stationary measure are = () ei .

The stationary measure does not depend on the two different quantum coins, that is, if + = , then, we
have the same measure. We should also remark that the stationary measure does not have an exponential
decay for the position. Furthermore, the defect and initial coin state strongly influence on the stationary
measure. In addition, the stationary measure is generally non-uniform for the position, however, if the
defect is defined by diagonal unitary matrix, then the stationary measure becomes a uniform measure.
4.2 3-state diagonal QW
Here we consider the 3-state space-homogeneous QW, whose time-evolution is described by the following
diagonal unitary matrix:
i
e
U = 0
0

0
ei
0

0
0,

(24)

where C is the determinant of U with | | = 1.

This is a special case of the 3-state diagonal QW with one defect. Now we show the stationary measure.
We obtain Theorem 3 in a similar way as Subsection 3.1, and we omit the proof here.
Theorem 3 The stationary measure of the QW is written by

||2 + ||2
(x 6= 0)
(x) =
,
||2 + ||2 + ||2 (x = 0)
where = L (0), = R (0), and = O (0). Remark that the eigenvalues are =

p
ei .

By putting a = ei , e = ei , i = , we also obtain the same outcome from Theorem 2. We emphasize that the stationary measure depends on position x, which is in marked contrast to that of 2-state
homogeneous case [14]. One of our basic future problems is to obtain the description of all the stationary
measure for the 3-state homogeneous QW by applying the method developed in Ref. [14].
5 summary
By using the SGF method [3, 7, 13], we obtained the stationary measure for 3-state homogeneous diagonal QW, 2 and 3-state diagonal QWs with one defect, and then, clarified the characteristic properties.
Through all the cases we saw, we also found that the eigenvalues are strongly influenced by the determinant
of the diagonal unitary matrices. From a viewpoint of classification, one of our essential future problems
is to apply the method [14] to the diagonal QWs we treated, and make clear the whole description of the
stationary measure. For instance, it is substantive to elucidate the conditions of the stationary measure
to be homogeneous, which may lead to classify the diagonal QWs with defects. As our results suggest, it
is precious to investigate the influence of the number of states and the diagonality of the defects on the
uniformity of stationary measure. Also, to develop the method to construct the stationary measure for
the QWs with plural defects is imperative. On the other side, it is interesting to consider the topological
invariants for the 3-state cases, which may lead to establish the interpretation for localization of QWs
from a viewpoint of condensed matter physics [4].

Acknowledgement
This work is supported by Toshiko Yuasa memorial fund, and the Grant-in-Aid for Scientific Research
(Challenging Exploratory Research) of Japan Society for the Promotion of Science (Grant No.15K13443).
References

1. A. Ambainis, Quantum walks and their algorithmic applications, International Journal of Quantum Information
1, 507-518 (2003)
2. M. J. Cantero, F. A. Grunbaum, L. Moral, and L. Velazquez, One-dimensional quantum walks with one defect,
Reviews In Mathematical Physics 24, 1250002 (2012)
3. T. Endo and N. Konno, The stationary measure of a space-inhomogeneous quantum walk on the line, Yokohama
Mathematical Journal 60, 33-47 (2014)
4. T. Endo, N. Konno, and H. Obuse, Relation between two-phase quantum walks and the topological invariant,
arXiv:1511.04230.
5. T. Endo, N. Konno, E. Segawa, and M. Takei, A one-dimensional Hadamard walk with one defect, Yokohama
Mathematical Journal 60, 49-90 (2014)
6. T. Endo and N. Konno, The time-averaged limit measure of the Wojcik model, Quantum Information and
Computation 15, 0105-0133 (2015)
7. S. Endo, T. Endo, N. Konno, E. Segawa, and M. Takei, Limit theorems of a two-phase quantum walk with
one-defect, Quantum Information and Computation 15, 1373-1396 (2015)
8. J. Ghosh, Simulating anderson localization via a quantum walk on a one-dimensional lattice of superconducting
qubits, Physical Review A 89, 022309 (2014)
9. J. Kempe, Quantum random walks, An introductory overview, Contemporary Physics 44, 307-327 (2003)
10. T. Kitagawa, M. S. Rudner, E. Berg, and E. Demler, Exploring topological phases with quantum walks, Physical
Review A 82, 033429 (2010)
11. N. Konno, Quantum walks;Lecture notes in mathematics, Springer 1954, 309-452 (2008)
12. N. Konno, The uniform measure for discrete-time quantum walks in one dimension, Quantum Information
Processing 13, 1103-1125 (2014)
13. N. Konno, T. Luczak, and E. Segawa, Limit measures of inhomogeneous discrete-time quantum walks in one
dimension, Quantum Information Processing 12, 33-53 (2013)
14. N. Konno and M. Takei, The non-uniform stationary measure for discrete-time quantum walks in one dimension,
Quantum Information and Computation 15, 1060-1075 (2015)
15. T. Oka, N. Konno, R. Arita, and H. Aoki, Breakdown of an electric-field driven system: a mapping to a quantum
walk, Physical Review Letter 94, 100602 (2005)
16. R. Portugal, Quantum walks and search algorithms, Springer, (2013)
17. A. Schreiber, K. N. Cassemiro, V. Poto
cek, A. G
abris, P. J. Mosley, E. Andersson, I. Jex, and Ch. Silberhorn,
Photons walking the line: A quantum walk with adjustable coin operations, Physical Review Letter 104, 050502
(2010)
18. N. Shenvi, J. Kempe, K. B. Whaley, Quantum random-walk search algorithm, Physical Review A 67, 052307
(2003)
19. S. E. Venegas-Andraca, Quantum walks for computer scientists, Synthesis Lectures on Quantum Computing
11-119 (2008)
20. S. E. Venegas-Andraca, Quantum walks: a comprehensive review, Quantum Information Processing 11, 10151106 (2012)
21. F. Z
ahringer, G. Kirchmair, R. Gerritsma, E. Solano, R. Blatt, and C. F. Roos, Realization of a quantum walk
with one and two trapped ions, Physical Review Letter 104, 100503 (2010)
22. A. Wojcik, T. Luczak, P. Kurzynski, A. Grudka, T. Gdala, and M. Bednarska-Bzdega, Trapping a particle of
a quantum walk on the line, Physical Review Letter A 85, 012329 (2012)